macro - discrimination index (part-whole-corrected)

[Alex Zimmerhofer]

Hi!

I'm working on a macro to calculate the discrimination index (part-whole
corrected) WITHOUT the command reliability... The macro should be adapted
easily for different data sets!

I don't know why - but it is not working (syntax error in vector dv)...

I'm at a loss...

THANKS

Alex


SET MPRINT=no.
DEFINE !partwhole(nby=!CMDEND).
VECTOR dv = i1 TO i7 /score_without_item(7).
COMPUTE sum_score_gesamt = 0.

* Calculating total score.
!DO !index1=1 !TO !nby.
- COMPUTE score_all = score_all + dv(!index1).
!DOEND.

!DO !index2=1 !TO !nby.

* part-whole-corrected total score.
COMPUTE score_without_item(!index2) = score_all - dv(!index2).

* correlation item-total score (corrected).
TITLE 'Discrimination index (Part-Whole correction)'.
CORRELATION dv(!index2) WITH score_without_item(!index2).

* correlation item-total score (without correction).
TITLE 'Discrimination index (without Part-Whole correction)'.
CORRELATION dv(!index2)  WITH score_all.
!DOEND.

!ENDDEFINE.

SET MPRINT=yes.
!partwhole nby=7.
EXECUTE.

[Raynald Levesque]

Hi

Try this untested code (if it does not work, please send me a sample data
file to facilitate debugging)

SET MPRINT=no.

*/////////////////////////////.
DEFINE !partwhole(nby=!CMDEND)
VECTOR dv = i1 TO !CONCAT('i',!nby) /score_without_item( !nby ).

COMPUTE score_all = SUM(i1 TO !CONCAT('i',!nby)) .

!DO !index2=1 !TO !nby
COMPUTE !CONCAT("score_without_item",!index2 ) = score_all - !CONCAT("dv",
!index2) .
TITLE 'Discrimination index (Part-Whole correction)'.
CORRELATION !CONCAT("dv", !index2) WITH !CONCAT("score_without_item",
!index2 ).
TITLE 'Discrimination index (without Part-Whole correction)'.
CORRELATION  !CONCAT("dv", !index2)  WITH score_all.
!DOEND

!ENDDEFINE.
*/////////////////////////////.

SET MPRINT=yes.
!partwhole nby=7.

Cheers!

Raynald Levesque [hidden email]
Visit my SPSS site: http://www.spsstools.net


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Alex Zimmerhofer
Sent: September 8, 2006 6:35 AM
To: [hidden email]
Subject: macro - discrimination index (part-whole-corrected)

Hi!

I'm working on a macro to calculate the discrimination index (part-whole
corrected) WITHOUT the command reliability... The macro should be adapted
easily for different data sets!

I don't know why - but it is not working (syntax error in vector dv)...

I'm at a loss...

THANKS

Alex


SET MPRINT=no.
DEFINE !partwhole(nby=!CMDEND).
VECTOR dv = i1 TO i7 /score_without_item(7).
COMPUTE sum_score_gesamt = 0.

* Calculating total score.
!DO !index1=1 !TO !nby.
- COMPUTE score_all = score_all + dv(!index1).
!DOEND.

!DO !index2=1 !TO !nby.

* part-whole-corrected total score.
COMPUTE score_without_item(!index2) = score_all - dv(!index2).

* correlation item-total score (corrected).
TITLE 'Discrimination index (Part-Whole correction)'.
CORRELATION dv(!index2) WITH score_without_item(!index2).

* correlation item-total score (without correction).
TITLE 'Discrimination index (without Part-Whole correction)'.
CORRELATION dv(!index2)  WITH score_all.
!DOEND.

!ENDDEFINE.

SET MPRINT=yes.
!partwhole nby=7.
EXECUTE.

[Alex Zimmerhofer]

Hi Raynald!

Thanks for your fast response. Great work!

But the macro isn't working (Text: dv1¬ SYNTAX ERROR--A subcommand or
variable name is expected here.  Perhaps a variable name is misspelled.
Valid subcommands for this procedure are VARIABLES, MISSING, PRINT, FORMAT,
MATRIX, and STATISTICS.¬ This command not executed.¬).

My data file contains 7 variables i1 to i7 (the item raw scores; each case
is the data set for one person)!


Thanks!

Alex